潜在狄利克雷分配模型在网络日志的应用

摘要

In recent years,the data mining technology based on the Web log is becoming a hot issue in the theoretical research and commercial applications,and the classification of Web users is one of the most important research in the field of data mining.Classification of Web users can help us find the similarities between the behavior of users,so as to set up the corresponding service project for specific user group.In this paper,according to the history information of the URL which users had accessed,we proposed a user classification method based on weighted potential Lejeune Dirichlet allocation (LDA) model,and the user is divided into different topic groups.Experiments show that this method can achieve very good classification results.%近年来,基于Web日志的数据挖掘技术逐渐成为理论研究和商业应用中的热点问题,而其中Web用户分类又是挖掘领域中最重要的研究主题之一.对Web用户分类能够发现用户之间相似的用户行为,从而针对具体用户群设置对应的服务项目.根据用户的历史访问网页地址(URL)信息,提出了基于加权潜在狄利克雷分配(LDA)模型的用户分类方法,将用户划分到不同的主题群体,实验表明,这种方法能达到很好的分类效果.